A Line-by-Line Analysis of the Harris-Allen op-ed in the National Law Journal Harris-Allen text The facts 1. Richard Sander of the University of California at Los Angeles School of Law asserts that affirmative action in law schools hurts black law students because it puts them in schools where their credentials are below the median; consequently, they cannot academically compete. Sander claims to have empirically proven this "mismatch thesis," but his findings have been contested. 1. There are 3 errors here: 1a. Just to be clear, the mismatch theory doesn’t say black students can’t compete; it says that any student (of any race) whose credentials are far below those of law school classmates tends to learn less than she would at a school where her credentials were closer to the median. When one adjusts for the mismatch effect, blacks achieve the same as everyone else. 1b. Sander has never claimed he has “empirically proven” the mismatch; it’s not a scientific way of speaking. Scientists and social scientists generally speak instead of theories being strongly supported or weakly supported by evidence. 2. Some researchers have documented a "reverse mismatch effect": Black law students attending higher-status schools do better, not worse, in terms of bar passage rates compared to counterparts attending lower-prestige schools. (This is true of all law students — higher expectations tend to improve student learning.) 2a. Those who find a “reverse mismatch” do so by ignoring serious selection bias problems and leaving law school grades out of their explanatory models, but it’s true there’s genuine controversy over the size and extent of the mismatch effect. 2b. If Harris & Allen believe their statement that “higher expectations tend to improve student learning”, then they should be leading the charge for the proposed study of California data, since that would be ideal data for proving their claim. 3. Others challenge Sander's findings because he does not pose a hypothesis and determine whether the evidence supports it, as social scientists do. Instead, he asserts conclusions and then scours for evidence to support his argument, as lawyers do. 3. This is a personal attack with no basis in reality. Consider, for example, Sander’s most recent article on the mismatch effect, “The Racial Paradox of the Corporate Law Firm.” The entire article is organized around articulating 5 distinct theories, considering all the available data, and evaluating each theory in terms of the evidence. Professor Ken Dau-Schmidt of Indiana University, who was the commentator at the July 2006 Law & Society meetings at a session that included Sander’s paper, singled out the degree to which Sander took all the alternative hypotheses seriously and did not argue beyond the available evidence. This is typical of Sander’s work. 4. Because [Sander] publishes without peer review — unlike social scientists — serious questions plague his research. 4. Another false personal attack. Sander’s Systemic Analysis of Affirmative Action article – the seminal piece in the current debate overmismatch effects – was actually peer-reviewed at two journals, the Stanford Law Review and the New York University Law Review, both of which asked prominent economists to assess the work. Indeed, of the 20-odd articles debating Sander’s mismatch effect that appeared in 2004, 2005, and 2006, it appears that Sander’s piece was the only one to be peer-reviewed by social scientists. This Harris-Allen attack is thus particularly ironic. 5. Sander now says there are data — the bar exam scores of blacks and Latinos in California — that can help resolve the debate. 5. Two inaccuracies here. (a) The relevant data is not just that of blacks and Latinos, but all bar-takers in the California data. Again, the mismatch theory contends that anyone whose credentials are well below those of her classmates can be affected. Many schools, for example, give preferences to older white students, and there is strong evidence that these students are affected by mismatch. (b) It’s not just “Sander” who says the data can resolve the debate. It’s also: Professor William Henderson, Indiana U.; Professor Vik Amar, UC Davis; Professor Doug Williams, chair of economics, University of the South; and Dr. Stephen Klein, psychometrician recently retired from RAND Corporation. These four joined with Sander in making the proposal to study the data. They have proposed a detailed study that has been endorsed by dozens of other empirical scholars, legal academics, law school deans, California lawyers, and the United States Civil Rights Commission. 6. Thus far, the bar examiners have denied his request. 6. A committee of the California Bar recommended in late June, in an unusual divided vote, that the bar study proposal not be approved. Their reasons, and a letter from Vik Amar and others questioning those reasons, can be found [here]. The Bar’s Board of Governors is taking up the issue on November 8th, 2007. 7. Sander asks: Why? Isn't it perfectly reasonable to get more data? Before we board the train, we must ask where it's going. Sander says the information will help because the bar scores — versus readily available bar passage rates 7. Two misconceptions here. (a) This is not primarily an issue of bar scores versus bar passage rate, but of aggregated data versus individual-level data. Currently, the only data made available by the bar is aggregate pass rates by racial group. It would be highly relevant and— "are a measure of what law graduates have actually learned; this allows one to study not only how the mismatch might affect bar passage but also how it might affect actual learning." useful to test the group’s various hypotheses using individual-level pass-fail outcomes, and in fact the team plans to do this. But actual scores contain far more information than simple pass-fail results. Any social scientists understands this. If one wants to study the relation between, say, weight and cholesterol, one can learn much more by knowing actual weights and cholesterol levels rather than simply knowing whether someone’s weight is over or under 170 lbs., and whether their cholesterol is over or under 200. (b) The quote is intended to highlight that there are two issues in the proposed study. One issue is whether the mismatch is partly responsible for the dramatically lower